Two instruments measuring the same night produce different numbers, and the difference is often larger than researchers expect. Sleep onset latency reported by a participant may differ from the actigraphic estimate by half an hour or more.
A digital sleep diary is frequently treated as the weaker of the two records, to be reconciled with the objective one wherever they disagree. That framing discards information.
A sleep diary for sleep studies measures a different construct, and the size and direction of the gap between the two is often the most interesting variable in the dataset. This connects directly to how continuous behavioral measurement is developing.
Two Instruments, Two Constructs
An actigraph infers sleep and wake from movement. A diary records what the participant experienced and remembers. These are not two measurements of one thing.
Terminology is worth settling early. Objective is a convenient shorthand for device derived, but actigraphy is itself an estimate produced by an algorithm, and describing it as ground truth overstates what the measurement supports.
Movement based scoring identifies quiescence. A participant lying awake and still is scored as asleep, which is why actigraphy tends to underestimate sleep onset latency and wake after sleep onset while overestimating sleep efficiency. The bias is systematic rather than random.
Self report captures perceived sleep, which is influenced by mood on waking, presleep arousal, expectation and the salience of remembered awakenings. It is a genuine measurement of something clinically meaningful, and in several contexts it predicts outcomes better than the objective record does.
The asymmetry runs in both directions. Actigraphy cannot detect a participant lying awake without moving, and a diary cannot detect a brief awakening the participant slept through. Neither instrument is failing when it misses what it was never designed to capture.
A review of objective sleep measures against self reported quality concluded that agreement depends heavily on which subjective construct is being assessed. Treating disagreement as measurement error assumes the two instruments should converge, and there is little reason to expect that they will.
This has a direct consequence for study design. Choosing a single primary sleep outcome commits the analysis to one construct, so the choice should follow from the research question rather than from whichever measurement is easier to collect.
Where the Numbers Diverge Most Predictably
The divergence is not uniform across variables, and knowing where it concentrates helps a study plan for it.
Sleep onset latency shows the largest and most consistent discrepancy, because the participant experiences the period before sleep as waking time while the device sees stillness. Wake after sleep onset follows a similar pattern for brief awakenings. Total sleep time usually agrees more closely, and sleep efficiency inherits the errors of both components.
Direction matters as much as magnitude. Some populations report worse sleep than the device records while others report better, and a study reporting only the absolute size of the discrepancy has discarded half the information.
Population matters. Older adults, participants with insomnia, and those taking sedating medication all show wider gaps than healthy young sleepers. Work on diary and actigraphy discrepancies in adults with fibromyalgia found that opioid dose predicted the magnitude of discrepancy and interacted with age to predict its direction.
The practical consequence is that a sleep diary and an actigraph will disagree more in exactly the clinical populations most studies are recruiting, so a protocol assuming close agreement is planning around a condition it will not meet.
Night to night variability adds another layer. A single night of disagreement tells you very little, while a consistent pattern across two weeks describes something stable about how a participant perceives their own sleep.
Reading Discrepancy as a Variable
Once the gap is expected, it can be measured deliberately rather than absorbed as noise.
Sleep state misperception is the clearest example. The difference between reported and measured total sleep time is a defined phenomenon with its own literature, and it distinguishes clinical groups from good sleepers reliably enough to serve as an outcome in its own right.
It also tracks treatment. Cognitive behavioral therapy for insomnia frequently improves subjective sleep before objective measures move, and a study collecting only one record will see either an early effect or no effect depending on which one it chose.
Reporting conventions have not caught up with this. Many papers still present diary and actigraphy results in parallel tables without formally examining the relationship between them, which leaves an analysis unperformed on data already collected.
Commentary on subjective and objective discrepancy during laboratory recording describes how sleep perception can be expressed as a ratio, which makes it straightforward to enter into a model alongside conventional variables. Defining that derived measure in advance is what turns a nuisance into a finding.
Specifying it in the protocol also avoids a common criticism. A discrepancy metric chosen after the results are visible invites the objection that it was selected to fit, whereas one defined in advance carries no such difficulty. Sample size planning follows the same logic, because a derived measure carries the variance of both underlying records.
What a Structured Diary Captures That Movement Cannot
Beyond perception, a diary records context that no accelerometer can infer.
Naps are the obvious case. A quiet afternoon hour is ambiguous in a movement record and unambiguous in a diary entry. Medication timing, caffeine, alcohol, illness, travel and shift changes are all invisible to the device and all capable of explaining an anomalous night.
Time in bed is a further example. The interval between getting into bed and getting out of it is reported directly and inferred only approximately from movement, and it serves as the denominator for sleep efficiency in most definitions.
Removals sit in the same category. A participant who reports taking the device off for two hours has converted a data gap into a documented interval, which is the difference between an excluded day and a partially analyzable one.
Subjective quality ratings round this out. They correlate imperfectly with every objective variable, which is precisely why they carry independent information. A study collecting both has two dependent variables available rather than one, and can report where they agree.
Environmental context can be captured here too. Questions about room temperature, noise disturbance or a change of sleeping location cost a participant seconds to answer and frequently explain nights that would otherwise be excluded as outliers.
Timestamping, Backfill and Diary Integrity
None of this holds if the diary itself is unreliable, and paper diaries frequently are.
The failure mode is well known. Entries are completed in batches shortly before a study visit, producing a record that looks complete and reflects recall across several days rather than daily observation. On paper the two are indistinguishable, because nothing records when the entry was made.
A digital instrument stores submission time as well as reported time. Blocking retrospective entry beyond a defined window goes further, guaranteeing that the dataset contains only entries made when they were supposed to be made.
Entry burden governs everything else. A daily form taking longer than about two minutes sees completion rates fall across a multi week protocol, so the item set has to be defended item by item rather than expanded by default. Language and accessibility deserve the same early attention.
Cohort level visibility completes the picture. Work comparing subjective and objective sleep measures in older adults found very little agreement between questionnaire assessment and wrist actigraphy, which is an argument for collecting both carefully rather than for trusting either alone.
Treating the Gap as Data
The instinct to reconcile two disagreeing records into one number is understandable and usually wrong. The disagreement carries signal, and averaging it away discards the part of the dataset that distinguishes participants from one another.
Pairing a wrist actigraphy device with a structured daily record gives a study two dependent variables and a derived third, which is a stronger position than either instrument produces alone.
Condor Instruments develops research grade actigraphs and our sleep diary app in Sao Paulo, with blocked backfilling and cohort level completion monitoring built in. Schedule a walkthrough with our team and we will show you how the two records align in practice.
Questions Researchers Ask About Combining Diaries and Actigraphy
Why do diary and actigraphy results differ?
They measure different constructs. Actigraphy infers sleep from movement and scores quiet wakefulness as sleep, while a diary records perceived sleep. The gap is systematic and largest for sleep onset latency and wake after sleep onset.
Should a study collect both?
In most cases yes. Each record contains information the other lacks, and the difference between them is itself a measurable variable that distinguishes clinical groups and tracks treatment response.
How is diary compliance monitored during a study?
A digital instrument records submission time for every entry and can display completion across an active cohort. That allows a coordinator to contact a participant on day four rather than discovering a gap at analysis.
Can diary entries be aligned to the actigraphy timeline?
Yes, provided both records share a timebase. Reported bedtime and rise time can then inform rest interval setting, and flagged nights can be treated separately rather than averaged into the cohort.
